Bryan Johnson's Longevity Claims, Checked Against the Papers
Forty-six claims from fifteen of his most-watched videos, pulled verbatim with timestamps and checked against the literature. Not one was flatly contradicted — he is more careful with other people's science than most of the field. The problem is somewhere else entirely.
The short version
I expected this to be the easy one. It was not.
Bryan Johnson cites real papers and mostly cites them correctly. His daily protein target — 1.6 g/kg — is precisely the breakpoint from the best available meta-analysis, which is a more defensible number than the one Peter Attia gives. He tells his audience the science on fasting "is not really there yet." He volunteers that the hair-growth compound he uses was only tested in mice. He summarises a 2025 creatine trial accurately including the part where creatine did nothing beyond the first week — a finding that cuts against a supplement his own company sells. Across 46 checked claims, zero were contradicted. That is a better record than anyone else in this series.
The problem is that the claims which make him famous are not the ones I could check. Fourteen of the forty-six have no citation attached — not because we could not find one, but because no source could exist. "Potentially the best sleep score in human history." "I've reversed my oral health age by over 20 years." "I think I've reduced my gray hair by about 30 to 50 percent." "I'm aging 50% slower."
These are measurements of one man's body, taken with instruments he owns, interpreted by people he employs, with no control and no independent verification. They are not false. They are unfalsifiable — and that is a stronger criticism than being wrong, because a wrong claim can be corrected and an unfalsifiable one cannot.
Where the science is sound
Being fair about this part is not padding — it is what makes the rest credible.
| Claim | The evidence |
|---|---|
| Daily protein target of 1.6 g/kg | Exactly the meta-analytic plateau above which added protein produced no further resistance-training gains. A well-chosen number. DOI |
| Morning 10,000-lux light "tells my body the day is beginning" | Standard, evidence-based circadian practice. DOI |
| Eyes-closed single-leg balance predicts all-cause mortality | Correct, and an unusually good pick — ten-second one-legged stance performance predicts survival in middle-aged and older adults. |
| Waist-to-height ratio beats BMI for cardiometabolic risk; 0.5 is the boundary | Supported by systematic review of 78 studies. DOI |
| Black pepper's piperine raises curcumin bioavailability | A classic, replicated pharmacokinetic finding. |
| Creatine did not increase muscle gain beyond training alone once wash-in was accounted for | An accurate reading of the 2025 trial — reported despite selling creatine. DOI |
| Strenuous evening exercise degrades sleep (4.3M nights, ~15,000 people) | Accurately described. The paper covers 14,689 individuals. DOI |
| "The science on fasting is not really there yet" | Correct, and notably more cautious than most of the field. |
The body-temperature problem
One claim is worth taking slowly, because it shows the whole pattern in a single sentence.
"Over the past couple years, my body temperature has dropped from 98.7 to 93.9. My metabolic systems have just become much more efficient."
93.9°F is 34.4°C. In any emergency department in the country that is clinical hypothermia, and it is a number that would prompt a workup, not a wellness claim. For scale: population body temperature in the US has fallen by roughly 0.03°C per decade since the 1860s — a real effect, and about one-sixtieth the size of the drop he is describing in himself over two years DOI.
He also gives the number three different ways. In this video it is a 4.8°F drop. In another it is "almost 4 degrees." In a third it is "roughly 3 degrees." A person who has genuinely undergone a remarkable physiological change knows what the change was.
The far likelier explanation is a measurement artifact — a peripheral or wearable sensor reading, not a core temperature — which would be unremarkable and worth nothing as content. "Efficiency" is an interpretation laid on top of an unexplained reading, and it is presented to millions of people as an established metabolic adaptation.
The claims that cannot be checked
| Claim | Why it cannot be checked |
|---|---|
| "Potentially the best sleep score in human history — 8 months of perfect sleep" | A proprietary wearable score. There is no registry of human sleep scores against which "in human history" could be evaluated. |
| "I've reversed my oral health age by over 20 years" | "Oral health age" is not a validated metric with year units. The underlying observation is his dentist's assessment of gum recession. |
| "I've reduced my gray hair by about 30 to 50 percent" | Self-estimate. No stated method, no baseline imagery. |
| "My lowest speed of aging was 0.48 … this roughly means I'm aging 50% slower" | DunedinPACE is a real instrument DOI — but it is validated on within-cohort trajectories, not as a personal, durable figure. Morgan Levine, who co-developed the advanced clocks, has said publicly that this conversion is not a valid use of the measure: clocks are not denominated in years. |
| 90 days of hyperbaric oxygen "lengthened my telomeres, reduced senescent cells, eliminated systematic inflammation, reduced a marker for Alzheimer's, rebuilt my microbiome" | The telomere and senescence portion echoes one small uncontrolled trial of about 30 people DOI. The rest — eliminated inflammation, an Alzheimer's marker, a rebuilt microbiome — has no published basis and is n=1 self-report presented as demonstrated effect. |
What his own data showed
This section is not my analysis. It is reporting by others, and I am relaying it with attribution because it bears directly on how much weight the personal claims above can carry.
In March 2025 The New York Times, working from internal Blueprint emails, spreadsheets and documents, reported that while Johnson publicly claimed in the Netflix documentary that his biological age had reversed by 5.1 years, internal blood-test charts covering January 2022 to February 2024 showed his biological age had increased by as much as ten years. The same reporting found that of roughly 1,700 participants in his self-experimentation study, about 60% experienced at least one side effect, with some participants' testosterone dropping and some becoming prediabetic on the diet plan.
Two of his own reversals belong here as well, and he deserves some credit for both:
- Rapamycin. In the documentary he described running "the most aggressive rapamycin protocol of anyone in the industry." In January 2025 he disclosed he had quit after about five years, saying the benefits did not justify side effects — recurrent skin and soft-tissue infections, lipid abnormalities, glucose elevations, raised resting heart rate — and citing a preprint indicating rapamycin accelerated aging across sixteen epigenetic clocks. On our tape he calls it "one of the most promising anti-aging drugs," which was a fair reading of the mouse record at the time. The mouse data is genuinely the strongest in the field. The human data was always his own.
- Plasma exchange. On tape, of the father-and-son plasma swap: "for my father it had significant results so it really was effective." He later reported no detectable benefit and discontinued it. The FDA had warned against young-plasma infusions for aging in 2019.
Reversing a public position on your own evidence is the right behaviour and I would rather see more of it. But it also means the confident version was broadcast to millions, and the retraction reached a fraction of them.
Two claims that are simply out of date
Aspirin. He takes 81 mg three times a week citing "some good evidence." For a healthy man in his forties, that evidence moved sharply the other way: the US Preventive Services Task Force's 2022 statement made primary-prevention aspirin an individualised decision for ages 40–59 and recommends against starting at 60 or older DOI. This is the one claim here I would actively counsel a patient about.
Acarbose "with metformin — there's good evidence that it extends lifespan." Half right, and the half that is wrong matters. In the NIA's Interventions Testing Program, acarbose did extend lifespan in male mice. Metformin, tested in the same programme, did not DOI. Bundling them into one sentence transfers acarbose's result onto a drug that failed the same test.
Grounded sheets
Small, but it is the clearest example of the failure mode. He says his bedsheets are grounded and "we found some good evidence on that." The most-cited paper in that literature is titled "Can electrons act as antioxidants? A review and commentary" — it is, as the title says, a commentary, not a clinical trial DOI. The grounding literature consists of small, low-quality, largely industry-linked studies. Someone spending two million dollars a year on measurement should be the least susceptible person alive to this, and the fact that he is not tells you the protocol is not filtered by evidence quality in the way the branding implies.
What I think this adds up to
Johnson is a better citer of science than most people in this space and a worse source of evidence about himself than almost anyone, and those two facts are not in tension — they are the same fact. The protocol's checkable parts are conventional and sound. The parts that justify the two million dollars are, by construction, not checkable by anyone but him.
If you are considering following any of this: the boring components are the evidenced ones. Protein at 1.6 g/kg, morning light, resistance training, fibre, olive oil, less alcohol, a real sleep schedule. Those are free or close to it and they are where essentially all of the demonstrated benefit lives. What the expensive tail buys is measurement, and measurement of one person without a control does not produce knowledge — it produces content.
How this was checked
Fifteen of Johnson's most-viewed videos were transcribed and each factual claim pulled verbatim with a timestamp. Each claim was matched to a specific paper and verified against the PubMed record — author, year, journal, finding — rather than against memory. Verdicts are supported, overstated, contradicted, or unverifiable, where unverifiable means no checkable source exists, not that the claim is false. Transcripts are machine-generated; garbled passages are noted. Reporting attributed to the New York Times is theirs, not ours, and is cited as reporting. Article metadata retrieved from PubMed. Full methodology →
This article is evidence appraisal, not medical advice, and does not establish a physician–patient relationship. Talk to your own clinician before starting or stopping anything discussed here — the aspirin section in particular is not a recommendation either way for you specifically.